We checked 254,854 Telegram channels. One in five is dead

TGList indexes 3.1M Telegram channels. For 254,854 of them we hold two timestamps: the date of the last post, and the date we looked. That pair answers a question nobody has answered at this scale — how much of Telegram still publishes anything.

19.2% had been silent between three months and a year. 47.9% had posted within thirty days. The median channel had said nothing for 36 days.

These 254,854 channels skew large. Their median subscriber count is 6,242, against 624 for the 2.1M channels we have not crawled deeply enough to measure. This article describes big, catalogued, findable channels. The tail below them is deader than what follows, not livelier.

Channels that pass two years of silence do not return

Size protects until a million subscribers, then stops

Death rates fall from the small bands up to 100k-1M, where 10.9% are dead. Above a million subscribers they double to 19.6%. The largest channels on the platform die at the rate of channels a hundred times smaller.

English channels die at twice the Russian rate

29.7% of English channels were dead, against 13.5% of Russian ones and 9.5% of Ukrainian ones.

Crypto is the deadest category and the largest by audience

Crypto carries 860.1M subscriptions, more than any other category, and 32.9% of its channels were dead against 7.1% for fashion. Two categories a technical audience is likely to be subscribed to sit near the top: techchannels at 32.1% and AI at 26.9%.

The 2024 tap-to-earn boom left 479 million subscriptions behind

1,131 channels hold at least 100,000 subscribers each and had not posted for over a year. Together they carry 478.9M subscriptions. 90 of them exceed a million subscribers apiece and hold 239.6M between them.

Crypto and gaming account for 64% of those abandoned subscriptions. English-language channels account for 76%. The tap-to-earn games of 2024 sit at the top of the list: Hamster Kombat reached 24.3 million subscribers and stopped, TapSwap 12.0 million, X Empire 10.0 million, Tomarket 8.4 million.

One episode explains three anomalies above. It is why the million-plus band dies at twice the rate of the band below it, why English is the deadest major language, and why crypto carries both the most subscriptions and one of the highest death rates. A speculative bubble gathered audiences larger than most national broadcasters and left them subscribed to channels that say nothing.

Posting cadence predicts death better than size does

Channels our analyser recorded as posting several times a day were dead in 9.2% of cases. Channels recorded as posting rarely were dead in 46.3% of cases. The gradient runs monotonically across all five levels.

Dead channels score three times higher on engagement

Our engagement rate divides views on recent posts by subscribers. It reads 0.45 for dead channels and 0.15 for active ones. By that metric the deadest channels on Telegram outperform the living ones threefold.

Views on a Telegram post keep accumulating for as long as the post exists, while the subscriber count of an abandoned channel decays. A post from 2024 has collected two years of views and divides by a shrinking denominator. If that accounts for the inversion, then engagement rate measures nothing on a channel that has stopped, and any ranking that rewards it promotes corpses. We have not separated the two effects. We publish the anomaly rather than a confident explanation of it.

Cut the data yourself

Pick a category, a language and a size band. The explorer recomputes the split for that slice and lists the largest channels in it that have been silent for over a year.

It runs on a contingency table, not on channel records. We ship counts per category × language × size band, and combinations holding fewer than five channels are suppressed entirely. The only channels named anywhere in the file are the 1,131 above 100,000 subscribers that had been silent for over a year — the subject of this article. Nothing about a live channel leaves our database.